TunaGAN: Interpretable GAN for Smart Editing
This work provides a tunable method for smart image editing, which is incremental as it extends existing GAN frameworks to improve interpretability and address mode collapse.
The authors tackled the problem of editing high-resolution face images using high-level user instructions by introducing TunaGAN, an interpretable GAN that builds on StyleGAN with an auxiliary network, achieving good qualitative and quantitative performance.
In this paper, we introduce a tunable generative adversary network (TunaGAN) that uses an auxiliary network on top of existing generator networks (Style-GAN) to modify high-resolution face images according to user's high-level instructions, with good qualitative and quantitative performance. To optimize for feature disentanglement, we also investigate two different latent space that could be traversed for modification. The problem of mode collapse is characterized in detail for model robustness. This work could be easily extended to content-aware image editor based on other GANs and provide insight on mode collapse problems in more general settings.